What range of rate-of-change are you looking at? Is your example (1 mA over 1 minute) typical, near the sensitivity limit, or just a guesstimate?

Taking 1mA over 1 minute - it's about 6% of full-scale, so with e.g. 10V full scale it works out at 10mV/second. That's going to be pretty tough with just analog stuff. A quick search for "low drift analogue sample-and-hold" comes up with figures 2 or 3 orders of magnitude worse!

JohnDG,

The 1 mA/minute is a guess, but probably not too far off.

If I read the 4-20 mA loop across a 250 ohm resistor (this is the largest resistance I can use), this gives me a 1-5 V output corresponding to 4-20 mA. 1 mA would be 250 mV. Sampling once/minute would be fine.

Is this a requirement for different loops, or just a single loop problem. If it's a flow loop, the problem is a bit harder because of the noise.

Your comment about sampling once a minute sounds like it may be a temperature loop.

Does it need to alarm on both rising and falling signals, or just rising signals?

Can you build a little piece of circuitry?

Dave R

It is a single loop. It is a gas detector, in which the presence of gas causes the 4-20 mA signal to increase in a cumulative fashion. As an example, if the device is exposed to a certain concentration for a certain period of time, the 4-20 mA output would increase from 8 to 9 mA. If the gas is removed, the output remains at 9 mA until the next exposure. If it were exposed to the same concentration for the same time period later on, the output would go from 9 to 10 mA etc.

Let's say that 100 ppm for 1 minute will cause this 1 mA increase. If this is the case, 1000 ppm will cause the output to rise 1 mA in 6 seconds. A 1 minute exposure would cause a 10 mA increase. So the slope of the current output is proportional to the concentration. I want to alarm at a certain concentration, so I want to trigger an alarm if a certain rate of change is exceeded.

The 4-20 mA will only increase, and I only need to alarm on an increasing rate of change.

The signal is fairly quiet, with less than .01 mA noise.

What is the range of concentrations that you would want to alarm on?

Dave R

DaveR,

The examples I have given are just that, examples. I don't have specific data yet, but 1 mA/minute is a good guess.

To answer your specific question, about 60 ppm.

OK, I have to ask - why no microcontroller?

Mike

I'm not really sure. It is a safety device for a nuclear installation, so I am assuming it is an NRC regulation. The customer has made it very clear that this is a firm regulation, no micro controller/processor and no software.

I don't know if it is because the radiation does bad things to micro controllers that it doesn't do to other things, or if software in general is a concern because it may have a "bug", and thus an unknown and untestable failure mode.

Sorry to put a downer on things, but I really don't think this is feasible using pure analog circuitry.

I could imagine a machine using perhaps hundreds of latching relays (or relays configured to latch) to serve as a memory of the values of the changing variable.

This would work, but A) would cost many thousands of pounds, B) would use a lot of power, C) would debatably be a digital system, D) would be practically impossible to analyse to give a meaningful Safety categorization - the MTTFd (mean time to fail to danger) would probably come out at about 10 minutes - and E) (consequence of (D)) - would need maintenance every 5 minutes.

I would suggest that you _{^respectfully} point out these things to the tendering authority, and propose a remotely located digital monitoring system.

I have very limited electronics knowledge, but I was very surprised at your post. I'm thinking that I may have not explained well enough.

To be clear, to be "analog" the device cannot contain a micro processor/controller or software. It can be "digital" beyond this requirement.

Ok, it is easy enough to have a circuit that gives a voltage out that is proportionate to the rate of change. That is, if the slope is 1 mA/minute, the circuit puts out 1 V. If the slope is 2 mA/minute, the output is 2 V etc. This is a simple RC circuit with an op amp to scale the voltage. This is simple, but I can post a circuit if needed.

Now, what I need is an analog (see above definition) circuit that will close a relay when a certain voltage is exceeded . In the above example, I need a circuit that will close a relay above a 1 V input. This is where I have problem. I would like an off the shelf "analog" device that can do this. I would like that the trigger voltage to be settable.

I can't believe that there was no way to do this before the microprocessor age.

"This is simple, but I can post a circuit if needed."

Please post (or give a link to) the circuit.

Here

You're right - I'd blinded myself with thoughts of another set-up. Sorry. I don't do enough electronics these days .

I did a quick spice simulation (R=1M, C=100μF, input 10mVs^-1, LT1001 op-amp) and got a good result - but it did show that the output's very sensitive to discontinuities/noise on the input.

So it seems you've answered your own question? Or were you after something more off-the-shelf?

Yes. I am looking for something off the shelf. I was hoping to find a single analog device that would detect the rate of change and close a relay. I assume that this operation is normally done in a DCS or similar. However, I can't be the first person to want to do this locally, and hoped that someone could direct me to a purchased module that would perform this operation.

I'm willing to come up with the differentiating circuit. I know the voltage sensing module is fairly common, I just did not know the name. Now I see what I'm looking for is called a limit alarm, voltage sensing alarm, voltage controlling relay, voltage sensing relay etc.

However, when I look at the off-the shelf modules, it is difficult to tell if any of them are analog.

I've used Broyce Control for current trip devices, and there was no indication of any digital processing (although I agree, one can't tell from the outside - or even from the datasheets).

I've asked them whether their 45050 voltage trip relay contains digital processing devices - I'll post their reply here.

I don't see why you can't use a conventional analog comparator. One with an open collector output could directly drive the relay coil (with appropriate protection diode etc.) You'd want to design the comparator circuit with hysteresis so the relay doesn't chatter but rather comes on and stays on, ignoring small ac noise on the input analog signal from your op amp scaler.

Because you have the scaling circuit to which to apply the variable setting, its output will be a known analog value when the "set point" rate is exceeded. You set the comparator's trip threshold to respond to that voltage. Thus the comparator section is not affected when the triggering rate is adjusted.

Here's an article from Analog Devices worth reading called "Curing Comparator Instability with Hysteresis":

http://www.analog.com/library/analogDialogue/archives/34-07/comparators/

Note that I have taken you at your word that the rate detection circuit will hold its value and rise with cumulative increase in gas concentration.

You can use an op amp as a comparator too, so if you have an unused section available it might be put to use here, if you can keep cross talk from influencing the comparator section.

If using that approach I'd let the op amp/comparator drive a separate transistor which activates the relay coil (or use something like a ULN2901-type device which is designed for inductive loads like the hammers in an impact print head; and has integral protection diode).

For further isolation, you could put an opto coupler between the detection/trip circuit and the relay coil.

Search for a voltage sensing relay (VSR). I've used them to prevent brown-out/memory loss in off grid supplied sites.

Tad. here's a couple of thoughts on your circuitry.

You are correct that the rate of change sensing needs to be an RC circuit with several seconds of time constant. The nature of the process change will effect the accuracy of your alarm point. If the process changes sufficiently to trigger the alarm and then diminishes to zero for some time, you would be OK. If the process change continues after the alarm at a slower rate, and then subsequently increases, the alarm would trip at a lower level, as the capacitor would not have reached it's charge from the previous increase.

How is the alarm reset? Should it turn off when the process rate diminishes below the alarm point?

You will need a means to establish the proper capacitor charge level on startup, or the alarm will activate as soon as the power is turned on.

What is the range of charge rates that must be accomodated?

Dave R

Pass the 4-20 mA through a resistor. Sense the voltage on the resistor on two different circuits. First circuit is an integrator with a large time constant (so its output will not vary much when your 1 mA rise occurs. Second circuit will determine that 1mA/minute rise. Make that voltage increase a bit larger than the output of the first integrator and compare them. The toggle will alarm about the increase.

The trick is to reconcile the 1 mA rise with a drift in the 4-20 mA signal.

Another useful analog approach. Take a look at the Linear Tech LT1637. In particular the application circuit shown at the top left and bottom right corners of page 13 (of 16) in the data sheet.

This part can measure the differential voltage (i.e., infer the loop current directly) on a resistor placed in your 4-20 mA loop. And it functions when the open circuit loop voltage is as high as 44 Vdc.

The application circuit would be followed by your variable gain op amp which is adjusted to set the triggering concentration rate. That output would feed the comparator circuit to drive the relay coil.

Or perhaps you'd introduce the adjustment in this stage to simplify the overall result.

Here's a link to the part's web page:

http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1010,C1778,C1764,P1766

Note what was said before about consideration of drift in the 4-20 mA loop current. You definitely need to analyze the various characteristics to be sure the resulting circuit will meet your specs and stay valid over operational variance.

Try a company called APCS. They have them in a DIN rail mount. If your request is very special they will make it for you.

Rok

Purchase a small PLC & do some programming... rather straight forward... & probably will work out cheaper in the long run..